List your company

Medical Question Answering on MedQA (Acc, ECE, Brier)

66.8AccuracyGRPO
45.79251.24656.762.154Jun 12, 2026

Evaluation results

MethodLinks
GRPOBackbone Model=Ministral 3 8B2026.0666.830.3762.66
SFTBackbone Model=Ministral 3 8B2026.0665.233.0667.83
BaseBackbone Model=Ministral 3 8B2026.066531.5665.73
CoRABackbone Model=Ministral 3 8B2026.0664.631.9165.41
CoRABackbone Model=Gemma 2 9B Instruct2026.0655.642.2286.48
GRPOBackbone Model=Gemma 2 9B Instruct2026.0654.643.2887.56
GRPOBackbone Model=Qwen2.5 7B2026.0654.435.1776.74
CoRABackbone Model=Qwen2.5 7B2026.065239.7284.79
SFTBackbone Model=Gemma 2 9B Instruct2026.065247.5894.83
BaseBackbone Model=Gemma 2 9B Instruct2026.0651.846.8693.86
SFTBackbone Model=Qwen2.5 7B2026.0650.444.4591.39
BaseBackbone Model=Qwen2.5 7B2026.0646.642.9493.13